Back to skill

Security audit

Prompt Reviewer

Security checks for vulnerabilities and agentic risk

Overview

This is a prompt-review helper with a disclosed Chinese output preference and no code execution, data access, persistence, or hidden behavior.

Installers should know this skill is designed to answer primarily in Chinese. If you need multilingual behavior, ask the maintainer to change the hard-coded language instruction to follow the user's requested language or make Chinese a default preference only.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (5)

Anti-Refusal Statement

High
Category
Anti-Refusal
Confidence
81% confidence
Finding

The phrase 'Always respond in Chinese' is an anti-refusal / unconditional compliance pattern because it imposes a non-negotiable behavior inside the skill. Such absolute directives can interfere with system or user instructions, normalize instruction override patterns, and make the agent less able to safely adapt when a different language, format, or safety behavior is required.

Content

Scanner excerpt · SKILL.md (reported line 43)May include surrounding context.

text

**CRITICAL INSTRUCTIONS:**
- Always respond in Chinese, as requested by the user's base rules.
- Be constructive and precise in your feedback.
- Ensure the optimized prompt is well-structured, clear, and robust.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The README specifies that the review output is 'primarily in Chinese' and shows section headers in Chinese, which imposes a language preference by default. The policy allows locale or language constraints only when the user is given a choice or when the constraint is clearly justified, neither of which is present here.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill hard-codes Chinese-only output regardless of the user's explicit language preference or the surrounding system context. While not directly enabling code execution or data exfiltration, this can override higher-priority user needs, reduce transparency, and create prompt-control behavior that weakens reliable instruction following.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

All natural-language examples and expected outputs are specified in Chinese, including required Chinese output terms, which may imply the skill expects or forces a specific language. The file does not indicate that language choice is optional or justified by a region-specific purpose, so this can violate language/locale flexibility expectations.

Content

No source excerpt is available for this finding.

Vague Triggers

Low
Category
Not specified by scanner
Confidence
87% confidence
Finding

This JSON manifest-like eval file contains generic activation-style example inputs such as "给我推荐几个好用的开源工具" and "写一篇关于中国历史的文章", which are broad everyday requests rather than narrowly scoped triggers. There are no explicit constraints, exclusions, or negative examples indicating when this skill should or should not apply, increasing the risk of unintended invocation if these examples are reused as trigger guidance.

Content

No source excerpt is available for this finding.